A 3d model of Complex A, La Venta, Mexico

Susan D. Gillespie a,n, Michael Volk b

a Department of Anthropology, University of Florida, P.O. Box 117305, Gainesville, FL 32611-7305 USAb Department of Landscape Architecture, University of Florida, P.O. Box 115704, Gainesville, FL 32611-5704 USA

a r t i c l e i n f o

Article history:Received 21 September 2013Received in revised form14 April 2014Accepted 27 June 2014Available online 5 July 2014

Keywords:La VentaOlmecArchival archeologyClay architecture3d Digital modelingDigitization

a b s t r a c t

La Venta was a large regional center located near the Gulf coast in Tabasco, Mexico. From ca. 800–400 BCit was the major Olmec capital in Mesoamerica. Despite its significance La Venta has received littlearcheological attention. The clay structures of its ritual precinct, Complex A, excavated in the 1940s–50s,were subsequently destroyed. Unfortunately, the published reports on those excavations are inadequate,with misleading archeological drawings. In order to obtain a more precise and comprehensive under-standing of La Venta the original excavation records were consulted, and field drawings and maps weredigitized to create more accurate 2d images as well as a 3d model of Complex A. This article summarizesthe process of digitizing the archival records and the interpretive benefits from utilizing 3d visualizationsof the site. Recounting the process may inform similar projects dependent on archival records when fieldmapping or excavation are no longer possible.

& 2014 Elsevier Ltd. All rights reserved.

1. Introduction

1.1. The importance of La Venta

La Venta was a major regional center of the Olmec culture inMesoamerica. Its apogee is dated to approximately 800–400 BC(uncalibrated), roughly coincident with the Middle Formativeperiod (Pool, 2007: 160). The site is located in the state of Tabascoin southern Mexico 15 km inland from the Gulf coast (GonzálezLauck, 1996: 73). First excavated in the 1940s La Venta became the“type site,” the basis for defining Olmec culture (Grove, 1997),especially its stone-working tradition and distinctive art style.Despite the absence of native stone on the coastal plain La Venta'sartisans crafted colossal sculptures out of boulders brought from theTuxtla Mountains some 100 km to the west, and finely made smallobjects of serpentine and jadeite, the latter material originatingnearly 500 km to the east (Diehl, 2004).

Although subsequent research in the Gulf coast area hasprovided more details on Olmec culture and chronology (Diehl,2004; Grove, 1997; Pool, 2007), as the regional capital La Ventaremains essential for understanding the Middle Formative Olmecs.Furthermore, the complex political and religious institutionsevident at La Venta have long been thought to have influencedsocietal developments in many other parts of Mesoamerica,including neighboring Maya peoples to the east (Coe, 1968;Drucker et al., 1959; González Lauck, 1996).

Once shrouded in tropical vegetation, the 1.5 km long civic-ceremonial core of La Venta has now been mapped, revealing over30 mounds and platforms. The site is dominated by a massiveearthen pyramid over 30 m high, perhaps the largest singlestructure at its time in Mesoamerica (González Lauck, 1988,1996; Pool, 2007: 157). Just north of the pyramid is a group ofclay platforms and small plazas designated Complex A. Excava-tions in Complex A first brought world attention to Olmec culture,revealing richly stocked stone “tombs” and numerous purposelyburied clusters of jade and other artifacts (“dedicatory offerings”)(Stirling and Stirling, 1942). The most enigmatic finds at La Ventawere three huge mosaic “pavements” all of the same design,crafted out of hundreds of polished rectangular blocks made fromimported serpentine. The mosaics, as well as two large deposits ofserpentine blocks not forming a design, were laid in great pits andthen immediately covered with clay fill. Nothing comparable tothese five “massive offerings” is known elsewhere in Mesoamerica(González Lauck, 1996: 78).

1.2. Excavations at La Venta Complex A

Although Complex A was small relative to the entire site, itremains the most thoroughly excavated and documented portionof La Venta's civic-ceremonial center. Three major excavationprojects were carried out in Complex A: in 1942 directed byMatthew Stirling and Philip Drucker (Drucker, 1952; Stirling andStirling, 1942); in 1943 directed by Stirling and Waldo Wedel(Drucker, 1952); and in 1955 directed by Drucker and RobertHeizer (Drucker et al., 1959; Drucker and Heizer, 1965, 1975). Thislast project was the most extensive and was dedicated to

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/daach

Digital Applications in Archaeology and Cultural Heritage

http://dx.doi.org/10.1016/j.daach.2014.06.0012212-0548/& 2014 Elsevier Ltd. All rights reserved.

n Corresponding author.E-mail addresses: sgillesp@ufl.edu (S.D. Gillespie), mikevolk@ufl.edu (M. Volk).

Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–81

investigating the construction history of the earthen platformsinto which the dedicatory and massive offerings were buried(Drucker et al., 1959: 1). Built out of specially prepared clays,the platform surfaces and the floor of the principal court wereregularly painted in colorful thin layers of clay and sand, much of itbrought to the site. At several intervals, great pits were dug tohouse the blocks of serpentine, after which the platforms wereenlarged and repainted.

While the architectural biographies of the individual structuresvary considerably, four construction phases (I–IV) were deter-mined for Complex A as a whole based on the sequence ofstructural modifications initiated by the massive serpentinedeposits (Drucker et al., 1959: 121–127; Drucker and Heizer,1965; González Lauck, 2007). In the absence of chronologicalinformation elsewhere from the site, these four constructionphases became the basis for the four “archeological” phases of LaVenta's occupation (Grove, 1997; Pool, 2007). The phases weretentatively dated by radiocarbon, but samples were not obtainedfrom every identified phase, leading to continuing questions aboutLa Venta's chronology (Grove, 1997: 72).

Soon after the 1955 project ended, the site was badly damagedby looting and development, especially Complex A (Drucker andHeizer, 1965), although some stratigraphic data were retrievedthere during a brief expedition in 1967 (Heizer et al., 1968a).Interpretations of La Venta's history and function, and by exten-sion of Olmec cultural practices, are thus dependent on theComplex A excavations published over a half-century ago. Espe-cially important is the 1955 instrument-made map of Complex A.As Heizer later observed, Complex A was “so torn up by bulldozersthat no surface feature whatsoever exists that can be identified asbeing present in 1955. The 1955 map of Complex A, therefore, isthe best we will ever have” (Heizer et al., 1968b: 139).

1.3. Shortcomings of the published excavation data

Unfortunately, the scant published excavation data are incom-plete and inadequate. The maps and profile drawings in particularare inaccurate and misrepresent the field data (Coe andStuckenrath, 1964). In the single major report of the 1955 fieldseason (Drucker et al., 1959) it is impossible to correlate thetextual descriptions of the excavations with the two-dimensionalprofile and plan views, many of which are schematic, leadingarcheologists to neglect the descriptive information in the text(Gillespie, 2011). Furthermore, the goal of the 1955 project was toreveal the four-phase history of construction of Complex A(Drucker and Heizer, 1965: 63) (Section 1.2). However, the variousarchitectural strata that were assigned to those phases “float” inthe published profile drawings because no datum was provided toanchor them in vertical space. Thus, the construction phases ofone clay platform cannot be correlated with the phases of anadjacent platform (Coe and Stuckenrath, 1964).

Heizer's reference (Section 1.2) to the “1955 map” is a singleplan map showing the structure footprints, excavation units, andlocations of offerings (Drucker et al., 1959, Fig. 4). This mapcompress all phases of the history of Complex A into one flattenedimage. Later maps of La Venta derived from the 1959 report omitimportant information or misrepresent the size or configurationof structures (Gillespie, 2011). Interpretations of Complex A havetended to emphasize its structural design, especially its notablebilateral symmetry (e.g., González Lauck, 1996: 76), although thatdesign is best evident only in the final phase of constructionshown in the single plan map. In sum, despite the site's impor-tance in Mesoamerican prehistory, archeologists, being so depen-dent on maps and drawings, cannot make much sense of the LaVenta excavations.

1.4. Reconstructing the architectural history of Complex A

The lack of adequate publication, especially by today's stan-dards, does not mean that archeologists can never know moreabout Complex A; neither does it imply that the excavationprojects themselves were poorly done or inadequately recorded.Although there are shortcomings in the published maps, theoriginal field maps and supporting data recorded during the1955 project still exist and provide the opportunity to create moreaccurate, computer-assisted images of Complex A.

In 2007 Gillespie began a project, “Reconstructing the Archi-tectural History of La Venta Complex A,” to create new profiles andplans primarily from the field records of the 1955 excavations. Theproject's research objective is to determine the sequence andtechnology of the ritual practices – the building and elaborationof the clay structures and placement of artifact caches – thatcreated and modified Complex A over time, similar to the inten-tions of Drucker and Heizer (1956) (Section 1.2). Volk was added tothe project for his expertise with digitization and visualizationsoftware and knowledge of landscape formations.

Because the surface architecture of Complex A is destroyed, it isnot possible to correlate the field records with the physical remainsat the site today. We therefore had to rely entirely on the availablearchived field notes and drawings, made when mapping instrumen-tation and recording standards were quite different. All of the imageshad to be digitized to create 2d profiles and plans at a single scalethat could then be developed into a 3d model of Complex A, tovisualize how it looked at different periods of its construction history.

1.5. Objectives for this article

The first objective is to summarize the digitization and modelingprocesses, detailing some of the problems, missteps, and successes(Sections 2 and 4). A second important goal is to briefly explain thegreat benefits of the 3d digital model (Section 3), which hastransformed our understanding of the site and its history of use(Section 4). We now have very different interpretations than thoseof the 1955 archeologists, who lacked the technology and means toutilize all their mapping data to evaluate the chronology andfunction of the Complex A ceremonial precinct.

2. Materials and methods

2.1. Major Complex A features

Complex A was a small group of low platform mounds andplazas approximately 13,520 m2 in extent located immediatelynorth of the 30 m tall pyramidal mound (Complex C) (Fig. 1). Itsfinished design plan is marked by bilateral symmetry – thearrangement of structures on either side of a virtual north–southcenterline 81 W of N (Drucker et al., 1959: 15). Bisecting thecenterline, from north to south, are the largest mound (A-2); arectangular plaza (Feature A-1, the Ceremonial Court) measuring2396 m2 and partially walled with adobe bricks and columnarbasalt pieces set vertically; the South-Central Platform within theCeremonial Court; and a platform (Mound A-3) just south of thecourt. The other structures form pairs positioned equidistant eastand west of the centerline: the Northeast and Northwest Platformswithin the Ceremonial Court, the Southeast and Southwest Plat-forms on the court's southern boundary, and Mounds A-4 and A-5flanking Mound A-3.

A trench excavated along the centerline from Mound A-2 tosouth of Mound A-3, begun in 1942 and extended in 1943,revealed a number of buried caches – the “dedicatory offerings”– including several presumed tombs and deposits of finely made

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–81 73

jade objects (Drucker, 1952). These finds further manifested thesignificance of the centerline in the complex's design plan. The five“massive offerings” placed in well-constructed pits were excavatedin 1943 and 1955, revealing that much of La Venta's architecture issubterranean. Three were composed of hundreds of workedserpentine slabs arranged in a mosaic pattern. Two mosaics wereburied in pits preceding the construction of the Southeast andSouthwest Platforms, with a third deposited just south of MoundA-3. Under the mosaic “pavement” in the Southwest Platform the1955 project discovered 28 stacked layers of serpentine blocks laidin a deep square pit; presumably a complementary massive stackexists under the Southeast Platform but it was not excavated. Twoother massive offerings discovered in 1955 were rectangular pitswhose bottoms were lined with one or more layers of serpentineblocks: Massive Offering 3 under the Ceremonial Court floor justsouth of Mound A-2 and Massive Offering 2 dug into Mound A-2itself. The massive offerings were believed to have initiatedconstruction phases II–IV.

2.2. Archived field records

For this project three archives were consulted: the RobertFleming Heizer Papers in the National Anthropological Archives(NAA) of the Smithsonian Institution in Suitland, Maryland; theRobert Fleming Heizer Papers in the Bancroft Library at theUniversity of California, Berkeley; and the Richard H. Stewartphotographs of the 1942 and 1943 La Venta projects at theNational Geographic Society's Image Collection in Washington,DC. Many of the photographs and slides from Stirling's andHeizer's La Venta projects are now accessible online at theSmithsonian's “Olmec Legacy” website http://anthropology.si.edu/olmec/english/imagesDB/index.htm (accessed 12.04.14).

The most important field records were from the 1955 project.They include nine instrument-made (alidade and plane table) maps,eight of them on oversize graph paper. These were the work ofRobert Squier, a University of California, Berkeley graduate studentwho had principal responsibility for mapping the site and excava-tions (Drucker et al., 1959: 4). Squier's “Drawing 2” provided ourcrucial base map, locating all the structures and some of the majorfinds of 1942–43 still in situ. His “Map 3” located the 1955 excavationpits and trenches at the same scale. Both maps pinpointed the samemapping station (Datum 2) as a singular reference point (Maps

Folder, R.F. Heizer Papers, NAA). Squier's maps formed the basis forthe published plan view of Complex A (Drucker et al., 1959, Fig. 4),which depicts most of the clay structures in terms of their footprintsor maximal horizontal extent. The field maps include Heizer'sadditions and corrections to Squier's original maps, and not all ofSquier's observations were included in the published map.

Other records consulted were the 1955 field journals of Heizer,Drucker, Squier, and Eduardo Contreras, a Mexican archeologist onthe project (Box 2, R.F. Heizer Papers, NAA). Additional resourceswere the color slides and some miscellaneous papers, includingloose drawings (Boxes 4, 19, 23, 24, R.F. Heizer Papers, NAA).Unfortunately, most of the slides were not labeled as to prove-nience, or were given a code designation we have not yetdeciphered. We were able to identify the locations in virtually allof them by comparison with the field drawings, although someslides had been labeled upside-down.

Most of the profile and plan views were in the field journals, ongraph or lined paper pages. With the exception of Contreras'sjournal, they were not drawn to scale but were rough-sketchedwith notated measurements of features and strata. To create theimages for the 1959 publication, in 1956 Contreras was given thedaunting task of converting the sketched profiles to scale drawings(Drucker et al., 1959: 3). We used the color excavation slides tohelp correct the shape and configuration of strata and features inthe original sketched drawings. However, slides of strata in thetrenches were not reliable for actual redrawing due to the angle ofthe photograph and potential distortion.

The most important new data in the field records were the over600 spot elevations recorded by Squier. Significantly, these eleva-tions were not used in the 1955 archeological interpretationsor to create the published profiles, which therefore “float” inspace. The elevations were critical to our project for aligning thetrench profiles in vertical space, filling in unexcavated areas whereno profile information was available, and comprehending thesequence of structural changes to the complex as a whole.

2.3. Digitizing the field maps and profile drawings

The challenge was to compile, organize, and digitize the variouspieces of archival information for each architectural feature andthen compile them into a single rendering of the entire complex.The first step was to scan notes, drawings, and slides and convert

Fig. 1. Map of Complex A architecture with the Pyramid (C-1) to the right (south) (based on Drucker et al., 1959, Fig. 3).

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–8174

them to jpeg, tiff, or Adobes Acrobats PDF files. For comparison,the published drawings from the 1942, 1943, 1955, and 1967projects were also scanned. The initial scans were not always toscale or in proportion, and were corrected in either AdobesPhotoshops 6.0 or AutoCADs 2010. All of this information wasthen organized according to the individual structure or feature,such as the Southwest Platform or the Northeast Entryway. Thenext step was to coordinate plan and profile drawings of the samestructures and features to check for any gaps or glitches indicatinga problem to be resolved (Section 4.2). The resulting drawingswere also compared with the color slides.

The 1955 project uniquely recorded relative elevations in Com-plex A from three instrument stations, recorded in Squier's fieldjournal, maps, and trench profiles. Conveniently, he standardizedthem all in reference to a single instrument station (Datum 2) tomake the readings consistent. However, he did not convert thestadia rod readings (mostly negative) into absolute elevations abovesea level. The stadia rod was marked with English measurementunits (feet and tenths of feet), and English measurements were alsoused in all the 1955 field records. For convenience our projectmaintained the English system. Following common practice, weassigned an arbitrary elevation of 100' to Datum 2 to convert the rodreadings to relative elevations. Because none of the featuresmapped in 1955 still existed unmodified by the 1960s (Section1.2), we cannot accurately convert our arbitrary elevations to actualelevations from the contour map made in the 1980s (GonzálezLauck, 1988).

2.4. Making the 2d model: ArcGISs and AutoCADs

As these data were being digitized and organized by individualstructure, the next task was to assess the best way to convert theoriginal archival data into updated 2d profiles and plans, andcompile them into a single rendering of the entire complex. Webegan by entering base point and line information for the site planinto ArcGISs (ESRI ArcMap 9.3s) using the Editor toolbar anddefining elevations for each feature. ArcGISs is a readily availablesoftware package typically used for mapping and data analyses.Line and point entry was based on Squier's Maps 2 and 3 (Section2.2), underlain and georeferenced in ArcMaps.

As a preliminary step, we then experimented with exportingthese data to ArcScenes to produce 3d visualizations, using TINS(triangular irregular networks) to define surfaces based on thepoints, lines, and elevations (Fig. 2). However, while ArcGISs iswell suited to processing and analyzing large amounts of complexdata, it is not as appropriate for entering detailed line and pointdata of the type we were using. It also produced what initiallyseemed to be an error that we could not trace, in that the base of

Mound A-2 was considerably lower than the bases of the otherstructures dated to construction Phase I (Section 1.2); this “error”is further discussed in Section 4.4. Due to the difficult andcumbersome nature of the process, we transitioned to a differenttype of software to continue our work.

Our solution was to export the GIS data into AutoCADs 2010 tocomplete our final updated 2d plans and profiles. Similar to theprocess used in GIS, we began by inserting and scaling rasterversions of the archival plans and profiles into CAD. Wherepossible, portions or all of the archival drawings were traced,while also referencing dimensions and notes from various sources.Because most archival renderings were not drawn to scale orproportion, tracing was not an option except for isolated portions.In these situations, we relied on the dimensions and descriptionsin various sources to create the drawing. We now have most ofComplex A in a single multilayered AutoCADs file; namely,Mound A-2 and the Ceremonial Court and its structures. This isthe area of the complex with the best stratigraphic data for thedifferent construction phases (Figs. 3 and 4).

2.5. Converting the 2d Files into a 3d model

Our final step was to export the 2d AutoCADs data (and any 3ddata that were created) into Googles Sketchups 6 to create adiagrammatic but still accurate 3d model. Sketchups is a freeprogram, extremely useful for fast and simple 3d modeling work.It presents disadvantages for complex models, both in terms of thecapabilities of the native tools and the actual ability of thesoftware to run and regenerate smoothly when working with adata-intensive model. It is less capable than AutoCADs whendealing with precise dimensions and forms. However, the lack ofprecision in the 3d model was less of an issue for our 3dvisualization and general interpretation requirements; precisionis maintained in the 2d renderings.

3. Results: 3d model of Complex A

3.1. Enhancing the 3d model

The resulting 3d model is based on a highly precise digital“skeleton” of strata, features, and embedded artifacts for visualizingthe complex strata of Mound A-2 and the Ceremonial Courtstructures. Textures and colors were added in Sketchups that,where feasible, are meant to resemble the original colorful sandsand clays that painted the structures, such as the brilliant yellowand purple coatings, and the adobe brickwork of the Southwest andSoutheast Platforms and the court wall (Fig. 5). We also imported

Fig. 2. Preliminary 3d image of Mound A-2 and the Ceremonial Court made in ArcScenes lacking the Southeast and Southwest Platforms.

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–81 75

jpegs of photographs of the major buried caches, in their appro-priate scales and positions, and manipulated images in Photoshopsto add background and shadow figures.

Quick Sketchups renderings of the southern structures (A-3,A-4, and A-5), which were less intensively excavated, were used tocomplete the 3d model of Complex A. For visual effect we alsoadded the pyramid (Mound C-1) immediately south of Complex A.That structure was digitized from a topographic map made in 1968(Heizer et al., 1968); its basal elevation was tentatively matchedwith that of Complex A. Like all other maps and drawings ofComplex A, our 3d model is anachronistic – the structures areshown at different periods of construction, in part because wecannot be sure of the appearance of every mound at the samepoint in time.

3.2. A virtual journey through Complex A

The 3d model allows us to virtually experience the solid formsof the structures and their relationships to one another, enhancedby a “fly-through” ⟨insert Complex A fly-over video clip⟩. Wecan recreate the visual experience of approaching the elevatedCeremonial Court from the south, the likely principal entrance.The view of the court between the Southeast and Southwest“bastion-like” platforms was blocked by the expansive South-Central Platform. The view from the other direction, north tosouth, puts the impressive great pyramid in the view shed (Fig. 6)(Video 1).

The fly-through also gives a sense of the relative elevations ofthe Complex A mounds. Importantly, the interior court platformswere only about knee-high, an experiential factor not appreciablefrom the 2d base map. We can also visualize the deep pits thatheld the massive offerings, marvel at how they could have beenstabilized in the sandy subsoil, and observe how this “inverse”architecture truly dwarfs the above-ground clay structures.Furthermore, the ca. 4 m thick adobe brick court wall, which hasbeen almost completely ignored in published maps and drawingsof Complex A (Gillespie, 2011), is revealed as an impressiveboundary sheltering ritual activities inside the court.

4. Discussion

This section details some of the missteps experienced inmaking the 3d model and the solutions we selected to resolvethe problems as they emerged from the process. Future research-ers will, and probably should, use a different procedure depending

on their circumstances. We also discuss the interpretive benefits ofthe 3d model in comprehending the construction history andritual use of Complex A.

4.1. Weighing software options

With regard to the three-step process (Section 2.4), we foundArcGISs too cumbersome and slow for our digitization goals. Itscapabilities to georeference and analyze multiple sets of complexdata were not necessary, and other software options are better for2d and 3d digitization. Thus, the initial GIS data were exportedinto AutoCADs. Although it is not as user-friendly as otherprograms, AutoCADs was readily available, supported by theuniversity, and familiar to the junior author. Importantly, weconstantly had to remind ourselves that AutoCADs permits moreprecision than was realistic or necessary for imaging earthenmounds originally recorded by imprecise hand measurements oralidade-based maps. In addition, because we were modelingearthen architecture, we found a strange void in the software'scapability: the solids modeling capabilities were too precise andcumbersome, and the terrain modeling capabilities were notfeasible due to the lack of adequate terrain-related information(i.e., contours).

All of our information was directly entered by hand. We brieflyconsidered raster-to-vector conversion software for automateddigitization, but the materials being digitized posed several con-straints. Not all the information was provided in a given scan, andmanual labor was required to delete unnecessary portions ofthe scan, add information that was incorrectly digitized, updatelineweights for printing, assign layers, and cross-reference dataderived from different sources. Furthermore, most of the scannedline and point information was not drawn to scale or in proportion(Section 2.2). There was no alternative than to enter that informa-tion manually, rectifying the various dimensions and spot eleva-tions and correcting the line drawings.

If archival drawings were available on plain paper with clearand accurate point and line information, software to automaticallydigitize that information would be worth exploring. The same istrue for text. All of the text was manually entered into ourdrawings, based on hand-written text in the drawing itself or infield journals, which in some cases was very difficult to read,utilized abandoned terminology, or required interpretation. Incontrast, clear and accurate text in archival records might bescanned and automatically entered via OCR scanning or similarsoftware.

Fig. 3. AutoCADs profile drawing of the north-south centerline trench running from Mound A-2 (right) through the South-Central Platform (left). This drawing is the key tothe individual trench profile details.

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–8176

4.2. Resolving other problems of digitizing archival records

File organization is always key and must be considered early inthe process. After a few missteps, we settled on a nomenclature forthe individual structural features and the complex-wide stratigra-phy. A well-labeled and separated system of layers within Auto-CADs was critical for us to isolate and view various featuresthrough time. Basic best practices for drafting, such as drawing at1:1 scale and judicious use of blocks and external references inAutoCADs, were also important and helpful.

Our first attempts to rectify the plan and profile views for theindividual structures (Fig. 7) sometimes revealed mistakes thatrequired correction – something had been either misrecorded inthe field or misinterpreted by us in attempting to fill in the gapsbetween trench profiles. We always searched for the origins of theerrors first in our own work, rather than in that of the fieldarcheologists. When it became clear that a mistake had been made

in the field (which was rare), we narrowed down the suspiciousreadings to the one that would cause the least repercussions forchanges in the relevant and adjacent drawings. The slides and the1940s photographs proved invaluable when deciding whether orwhere a mistake had been made.

It was also challenging to line up the various structures andfeatures horizontally with one another according to their relativepositions on the base map (Section 2.2). Errors were revealed inthis process that otherwise would have remained hidden. Fortu-nately, we had the advantage of being able to align strata verticallyacross the complex. The spot elevations were sufficient to create astandard z-axis for the area from Mound A-2 south to Mound A-3and including Mound A-5. Our profiles were finally anchoredin space.

Furthermore, we were able for the first time to situate the 1967excavation trenches (Heizer et al., 1968a) within the 1955 map,and to line up the stratigraphic information from the 1955 and

Fig. 4. AutoCADs plan view of Complex A from Mound A-2 to the Southwest and Southeast Platforms. Excavated areas (1942/43, 1955, and 1967) are in white.

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–81 77

Fig. 5. 3d View of the northern half of Complex A from the south, with added vegetation and shadow figures for scale.

Fig. 6. A video still shows the great pyramid (Mound C-1) looming behind the Complex A structures; view from the north looking south. The pit for Massive Offering 2 inMound A-2 is shown in the foreground.

Video 1. Complex A fly-over video clip: 3d flyover of Complex A, with the Complex C pyramid in the background. A video clip is available online. Supplementary materialrelated to this article can be found online at http://dx.doi.org/10.1016/j.daach.2014.06.001.

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–8178

1967 projects in three-dimensional space (see Fig. 4, Section 2.4).Those latter excavations provided crucial information on thefluctuating northern extent of Mound A-2, something not recordedin 1943 or 1955.

4.3. Rethinking the history and design plan of Complex A

A major goal of the 1955 excavators was to re-create theconstruction history of Complex A and its individual structures.Unfortunately they did not integrate the vertical positioning of thestructures and features in their interpretations, although thosedata were available. They were thus limited to demonstrating fourmajor construction phases for the Ceremonial Court based on asequential series of court-wide floors (Section 1.2). One of theirprincipal conclusions was that a formal design plan of bilateralsymmetry was maintained from Phase I to Phase IV, indicative ofan uninterrupted Olmec or “La Venta” occupation (Drucker andHeizer, 1965: 64; Drucker et al., 1959: 14). However, their excava-tions also revealed portions of several painted low clay platforms

beneath the Phase I floor layers, which they concluded dated to a“pre-La Venta” occupation (Drucker et al., 1959: 124).

Nevertheless the 1955 trench profiles published in 1959 did notsupport the interpretation of a single design plan maintained sincethe origin of the building group. For example, the NortheastPlatform seemed to predate the Northwest Platform by an entireconstruction phase, in that the Phase I floor layers abutted the baseof the Northeast Platform, while the Northwest Platform was builtatop that floor (Coe and Stuckenrath, 1964: 6; Gillespie, 2008).Furthermore, although bilateral symmetry is readily apparent inthe footprints of structures on the 1955 base map (Fig. 1), ourreconstructed plan views based on the profile drawings (Fig. 4)reveal that even the twinned Northeast and Northwest Platformswere subjected to quite different ritual purposes and were notidentical in appearance or use (Gillespie, 2011: 25–26).

With the 3d model we can now show change through time forindividual structures and the complex as a whole. The differentlayers of the AutoCADs file (Section 2.4) reference major andminor construction phases. They can be turned on and off,

Fig. 7. Aligning profile and plan views of Mound A-2 (right) and Massive Offering 3 (left) in the Ceremonial Court.

Fig. 8. A video still shows the Phase II massive offerings and mosaic “pavements” in the Southwest and Southeast Platforms. View looking north.

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–81 79

rendering images of the complex at various times in its history.A video animation ⟨insert link to Complex A construction historyanimation⟩ demonstrates the changes to the northern half ofComplex A through the major construction phases, including theexcavation of large pits to hold the massive offerings of theSouthwest and Southeast Platforms (Fig. 8). Our modeling alsoincorporates the buildup of the court floors and additions to theheights of the mounds in vertical space, resulting in a verydifferent interpretation of the construction history and designplan than is evident in the published maps (Video 2).

4.4. The importance of Mound A-2

Drucker and Heizer (1965: 39–40) insisted that the CeremonialCourt (designated A-1) was always the “principal structuralfeature” of Complex A, even as they noted that “particularattention was given throughout the entire span of time” to MoundA-2 (Drucker et al., 1959: 34). Mound A-2 was the tallest structurein Complex A and had the most complicated stratigraphic history.Our 2d and 3d models also indicate that it was the oldest and mostimportant from the very beginning of its existence.

The making of what would become Mound A-2 started withsequential deposits of thin layers of sand upon undisturbedground. Mound A-2 became a more proper flat-topped platform,about 0.3 m tall, with a deposit of prepared clay atop the earliersand layers (stratum j-3 in Drucker et al., 1959: Fig. 10). Dedicatorydeposits of vari-colored sands were gradually laid abutting thesouth edge of Mound A-2, eventually reaching its summit. Then asecond, much larger stepped platform of prepared red and whitesandy clays, over 1 m thick, was erected over the first phase(stratum i-1 in Drucker et al., 1959: Fig. 10). Drucker and Heizerconsidered the i-1 clay platform to date to Phase II of theCeremonial Court (Drucker et al. 1959: 46), but our z-axis indicatesthat at this stage Mound A-2 still pre-existed all other structuresand the Phase I floors. This is why our initial GIS model showedthe base of A-2 was lower than the bases of the other platforms,something we thought at the time was an error (Section 2.4).

After or in concert with this major rebuilding of A-2, brownsand was brought in to begin to level the area immediately southof the platform, which would become the Ceremonial Court. On

top of that fill at least three small low clay platforms were erectedin the general vicinity of the later Northeast, Northwest, andSouth-Central Platforms (Fig. 9). Additional fill to further raisethe lower, western portion of the court area buried the nascentNorthwest and South-Central Platforms, which is why the 1955excavators found remnants of “pre-Phase I” structures under them.This also explains why the Northeast Platform stratigraphicallyseemed to predate the Northwest Platform (Section 4.3), whoseearlier construction phase had been covered by the sandy fill.

4.5. A new interpretation of Complex A

Taking into account the 3d model of the construction changesto the structures and the complex as a whole at a more refinedtemporal scale, the ability to locate precisely the various dedica-tory and massive offerings, the incorporation of otherwise invi-sible ritual acts such as digging pits into the platforms, and theattention to building materials, we offer a different interpretationof the function and meaning of Complex A. From the beginning,this ritual precinct was devoted to the sacred locale of Mound A-2,a focus of ritual deposition and extraction of materials for genera-tions. The smaller platforms, eventually enclosed in a walled court,separated off the southern edge of Mound A-2 in particular as aplace for intense and continuous devotional activities. The adobebrick Southeast and Southwest Platforms were actually part of theadobe brick walled enclosure, erected at the same time, and arebetter considered “bastions,” as they were labeled in the 1955field notes.

The largest Massive Offering in areal extent was dug throughthe court floor just south of Mound A-2, initiating constructionPhase III. Massive Offering 3 was the locus for several dedicatoryofferings positioned above the six layers of serpentine blocksplaced at the bottom of the great pit. This massive deposit,abutting Mound A-2, was a sacred place within the already sacredlocale of the Ceremonial Court. Phase III also witnessed the shifttowards placing offerings along the north-south centerline, espe-cially above Massive Offering 3. The centerline per se seems not tohave been important until Phases III and IV, indicating a change inthe plan and use of the complex that nevertheless kept the focusof ritual attention on Mound A-2.

Video 2. Complex A construction history animation: video animation of changes to the Complex A architecture over the history of its active use. A video clip is availableonline. Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.daach.2014.06.001.

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–8180

Finally, in Phase IV a massive offering pit was dug into thecenter of Mound A-2 itself (Fig. 6), at the bottom of which a singlelayer of serpentine blocks was laid. “Tombs” were located at themound's summit directly over that serpentine deposit. The center-line orientation is most evident in Phase IV in terms of thelocations of buried offerings, but for the first time they wereplaced only outside of the Ceremonial Court, which was a majorshift in the ritual use of this precinct (Gillespie, 2008). Also inPhase IV the parallel structures A-4 and A-5 were erected, furthermarking and blocking access to the southern portion of Complex Abetween the Ceremonial Court and the pyramid.

5. Conclusion

Aside from the specific new interpretations regarding ComplexA derived from the 3d model (Sections 4.4 and 4.5), more generalconclusions can be reached regarding the significance and replic-able utility of this project. The first is the potential for digitizingarchival archeological records to create a data-base for 2d and 3dvisualizations. Although it can be difficult and time-consuming toretroactively craft maps and drawings in this way, this possibilityshould be considered for the numerous sites at which archeolo-gical research was conducted in the past but where on-site digitalmapping using new technologies is no longer possible due tothe subsequent destruction of features or the inability to obtainpermission from land-owners or authorities for such research.

Secondly, this project highlights the importance of archivingfield records in repositories where they can be made available toresearchers. Gillespie's one-day exploratory trip to the NationalAnthropological Archives in 2005, to discover whether the originalLa Venta field records deposited there by Heizer held any unutilizeddata to improve on the published maps, turned up a surprisingamount of information. Finally, the authors' collaboration, combin-ing different expertises (archeology and landscape architecture), hasproven very productive, yielding more complex and useful resultsthan a single disciplinary focus might have achieved.

Acknowledgments

Vyrtis Thomas, Archives Specialist, and Catherine O'Sullivan,Reference Archivist, of the NAA provided assistance in the con-sultation of the Robert Fleming Heizer Papers, and Bill Bonner,facilitated consultation of the NGS Image Collection. That 2007

research was funded in part by the Foundation for the Advance-ment of Mesoamerican Studies, Inc. (FAMSI) Grant #07054,awarded to Gillespie; FAMSI funds were not involved in thewriting of this article. Joshua R. Toney produced the GIS data forComplex A. Logistical support for this project was provided byChristian Russell of the University of Florida Land Use andEnvironmental Change Institute.

References

Coe, M.D., 1968. America's first civilization: Discovering the Olmec. AmericanHeritage Publishing, New York.

Coe, W.R., Stuckenrath Jr., R., 1964. A review of La Venta, Tabasco, and its relevanceto the Olmec problem. Kroeber Anthropological Society Papers, 31. Universityof California, Berkeley, CA, pp. 1–43.

Diehl, R.A., 2004. The Olmecs: America's first civilization. Thames and Hudson,London.

Drucker, P., 1952. La Venta, Tabasco: A study of Olmec ceramics and art. Bureau ofAmerican Ethnology Bulletin, 153. US Government Printing Office, Washington,DC.

Drucker, P., Heizer, R.F., 1956. Gifts for the jaguar god. Natl. Geogr. Mag. 110,366–375.

Drucker, P., Heizer, R.F., 1965. Commentary on W. R. Coe and Robert Stuckenrath'sreview of Excavations at La Venta, Tabasco, 1955. Kroeber AnthropologicalSociety Papers, 33. University of California, Berkeley, CA, pp. 37–69.

Drucker, P., Heizer, R.F., 1975. Archeological investigation of the site of La Venta,Tabasco, Mexico. In: Oehser, P. (Ed.), National Geographic Society ResearchReports 1890–1954. National Geographic Society, Washington, DC, pp. 387–394.

Drucker, P., Heizer, R.F., Squier, R.J., 1959. Excavations at La Venta, Tabasco, 1955.Bureau of American Ethnology Bulletin, 170. US Government Printing Office,Washington, DC.

Gillespie, S.D., 2008. History in practice: ritual deposition at La Venta Complex A.In: Mills, B.J., Walker, W.H. (Eds.), Memory Work: Archaeologies of MaterialPractices. School for Advanced Research, Santa Fe, NM, pp. 109–136.

Gillespie, S.D., 2011. Archaeological drawings as re-presentations: the maps ofComplex A, La Venta, Mexico. Latin Am. Antiquity 22, 3–36.

González Lauck, R.B., 1988. Proyecto arqueológica La Venta. Arqueología 4, 121–165.González Lauck, R.B., 1996. La Venta: an Olmec capital. In: Benson, E.P., de la Fuente,

B. (Eds.), Olmec art of ancient Mexico. National Gallery of Art, Washington, DC,pp. 73–81.

González Lauck, R.B., 2007. El Complejo A, La Venta, Tabasco. ArqueologíaMexicana. 15 (Sep–Oct), no. 87, pp. 49–54.

Grove, D.C., 1997. Olmec archeology: a half century of research and its accomplish-ments. J. World Prehist. 11, 51–101.

Heizer, R.F., Drucker, P., Graham J.A., 1968a. Investigations at La Venta, 1967.Contributions of the University of California Archaeological Research Facility,Berkeley, CA, 5, pp. 1–33.

Heizer, R.F., Graham, J.A., Napton, L.K., 1968b. The 1968 investigations at La Venta.Contributions of the University of California Archaeological Research Facility,Berkeley, CA, 5, pp. 127–153.

Pool, C.A., 2007. Olmec archaeology and early Mesoamerica. Cambridge UniversityPress, Cambridge.

Stirling, M.W., Stirling, M., 1942. Finding jewels of jade in a Mexican swamp. Natl.Geogr. Mag. 82, 635–661.

Fig. 9. A “pre-Phase I” image shows Mound A-2 as a large stepped clay platform, along with the mapped fragments of early versions of the Northwest, Northeast, and South-Central Platforms. The dimensions of the latter three small platforms are not known.

S.D. Gillespie, M. Volk / Digital Applications in Archaeology and Cultural Heritage 1 (2014) 72–81 81